Skip to main content
U.S. flag

An official website of the United States government

Publication Date
14 September 2016

Insights from a Refined Decomposition of Cloud Feedbacks

A refined decomposition better connects cloud feedbacks to individual processes and more clearly reveals robust feedbacks.
Print / PDF
Powerpoint Slide

Scientists in the Cloud Processes Research Group at Lawrence Livermore National Laboratory quantified changes in Earth’s energy budget due to changes in cloud properties that accompany global warming in climate models, known as cloud feedbacks.  By distinguishing between feedbacks due to low clouds that reside in the boundary layer from feedbacks due to non-low clouds in the free troposphere, they were able to better attribute cloud feedbacks to individual processes, to highlight feedbacks that are robust in sign across models, to quantify sources of inter-model spread in cloud feedback, and to estimate a “null hypothesis” climate sensitivity from well-understood and robustly-simulated feedbacks.


Despite a large inter-model spread in simulated cloud feedback, the refined decomposition reveals that models systematically simulate positive feedbacks from decreasing low cloud amount and increasing high cloud altitude and negative feedbacks from increasing low cloud optical depth.  Whereas recent studies indicate that the negative optical depth feedback is overstated in models, the robust positive feedbacks are well-supported by observations, theory, and high resolution modeling.  Including these two positive cloud feedbacks, the “null hypothesis” climate sensitivity arising from well-understood and robustly-simulated feedbacks is at least 3 degrees Celsius.


Cloud feedback – the change in Earth’s energy budget due to warming-induced changes in cloud properties – represents the largest source of inter-model differences in the amount of global warming per doubling of carbon dioxide. Decomposing cloud feedback into components due to changes in cloud amount, altitude, and albedo provides valuable insights into its physical causes. In this study the researchers quantified feedbacks from changes in these three cloud properties separately for low boundary layer clouds and free tropospheric clouds. It revealed that all climate models simulate positive feedbacks from increasing free tropospheric cloud altitude and decreasing low cloud cover and negative feedbacks from increasing low cloud albedo. Low cloud amount feedback is the dominant contributor to spread in net cloud feedback, but its anti-correlation with other components damps overall spread. Compared to the conventional cloud altitude feedback computed considering all clouds collectively, the ensemble mean free tropospheric cloud altitude feedback is 40% smaller and is better constrained. The authors estimated the “null hypothesis” climate sensitivity arising from well-understood and robustly-simulated feedbacks to be at least 3 degrees Celsius.

Point of Contact
Mark Zelinka
Lawrence Livermore National Laboratory (LLNL)
Funding Program Area(s)